Load Model from a GCP Cloud Function about greykite

[KaixuYang]

Hi @kenzie-q I'm not familiar with GCP or cloud function. What are the challenges? Is it a linux environment? Would you be able to use the current functions to write to local storage, compress into tar, then upload to cloud storage?
Similarly for loading, get the tar file, unpack it and read it. Does this work for you?

[kenzie-q]

Hi @KaixuYang ,

So a cloud function on GCP has a max run time of 10 minutes, and there is no local memory. Therefore we can't use the existing code to read in the model from cloud storage. Writing the model to Cloud storage isn't the issues. I was able to write a function that read all the files and directories to storage. I was also able to edit the existing load function to read from storage, however, it takes longer than 10 minutes to run unfortunately. I'm not sure if there is a way to speed up the process. I'm trying to commit the existing code I have to this repo to be of assistance.

[kenzie-q]

The code edit's I've made for overall functionality are in PR #45

[KaixuYang]

Hi @kenzie-q Thank you for the detailed info and code contribution. It's great that you find a way to make it work for the GCP use case. However, on one hand, we feed that this change slightly deviates the library's roadmap, and on the other hand we won't be able to test throughly the functionality and maintain it at this time, so we are sorry that we won't be able to accept this pull request. You are welcome to fork the Greykite repo and use the functionality on your use case. Feel free to open a new issue if you have more questions!

Best,
Kaixu